Six trials (n=407) were included in the review. Sample sizes ranged from 20 to 109 patients. Trial quality was variable. Scores ranged from 3 to 8 out of 10. Few trials adequately concealed allocation or used blinding of clinicians or assessors.
Short-term (six to seven weeks): Compared with physiotherapy, corticosteroid injections had a statistically significantly greater improvement in shoulder disability (SMD 0.74, 95% CI 0.45 to 1.04, I2=0%; three trials) and passive external rotation (SMD 0.66, 95% CI 0.22 to 1.10, I2=48%; three trials). There was no significant difference between physiotherapy and corticosteroid injections in terms of pain (as measured by the visual analogue scale).
Mid-term (12 to 16 weeks): Compared with physiotherapy, corticosteroid injections had a statistically significantly greater improvement in shoulder disability (SMD 0.32, 95% CI 0.02 to 0.62, I2 = 0%; three trials). There was no significant difference between physiotherapy and corticosteroid injections in terms of pain (as measured by the visual analogue scale) or passive external rotation.
Longer-term (26 to 52 weeks): Compared with physiotherapy, corticosteroid injections had a statistically significantly greater improvement in shoulder disability (SMD 0.36, 95% CI 0.03 to 0.69; two trials) and passive external rotation (SMD 0.44, 95% CI 0.11 to 0.77; two trials). There was no significant difference between physiotherapy and corticosteroid injections at 26 weeks in terms of pain (as measured by the visual analogue scale). At 52 weeks, there was no significant difference between physiotherapy and corticosteroid injections in terms of shoulder disability, but there was a small benefit in terms of pain (SMD 0.36, 95% CI 0.02 to 0.70; two trials).